3. CHEM. SOC., CHEM. COMMUN., 1983 Reversible Intercalation of Graphite b a 103 y Fluorine: a New Bifluoride, C,,HF,, and Graphite Fluorides, C,F (5 > x > 2) Thomas Mallouk and Neil Bartlett* Department of Chemistry, University of California, and the Materials and Molecular Research Division, Lawrence Berkeley Laboratory, Berkeley, California 94720, U.S.A, A blue-black first-stage graphite fluoride, C.F (5 > x > 2), in which the planar carbon-atom sheets of graphite are preserved, is made by intercalation of graphite with fluorine in the presence of liquid hydrogen fluoride, at ca. 20 °C, the oxidation (which is electrochemically reversible for x > 2.6) proceeding via a highly conductive second stage salt: C,2tHFo~. The recent discovery? in these Jaboratories that the intercala- tion of graphite by AsF,, in the presence of elemental fluorine, resulted in the incorporation of ‘extra’ fluorine to a limiting composition C,AsF,.}F, has prompted an examination of the factors which govern the intercalation of graphite by fluorine itself. It is common knowledge that graphite is not intercalated spontaneously by fluorine at room temperature and ordinary pressures, but that interaction between these elements docs occuc at higher temperatures (usually ca. 400 °C) to yield fluorides such as (CF), and (C;F),,, in which the F atoms are covalently bound to tetrahedral carbon? H owever, more than thirty years ago, Rudorff and Ridorff' had discovered that graphite does intercalate fluorine spontancously, at room temperature, in the presence of gaseous hydrogen fluoride to yield a first-stage graphite fluoride C,F with x values ranging from 3.57 to 4.03 which they termed ‘tetracarbon mono- fluoride’. X-Ray diffraction data showed ‘hat the carbon-atom layers in their C,F materials were planar and that the F ligands were covalently bound to carbon atoms. Our work has shown that the Riidorfis’ species is part of a more extensive system of true intercalation compounds including hydro- fluorides, a second-stage bifluoride C,HF,, and a first-stage compound C,F (5 > x > 2). For much of this composition range the intercalation has proved to be reversible. The materials are sufficiently conductive to facilitate reversible clec- trochemical oxidation or reduction over the range C,..F to graphite. The intercalation compounds are synthesised princi- pally by using anhydrous HF and F, (2 atm), or alternatively, using electrochemistry to ca. C.F, the latter method being impracticable for x < 2.6 owing to their very low conductivity. With anhydrous fiquid HF and gaseous F, (2 atm), at 20°C, graphitet intercalation to the second-stage salt, of approxi- mate composition C,;+HF,~, occurs within two hours. A third- Stage salt has been observed as a major phase after only several minutcs of interaction. A less impure sample of the t Union Carbide SP 1, or powdered HOPG interacts in the time stated. For larger HOPG pieces the required interaction times increase markedly with size. 104 J. CHEM, SOC., CHEM. COMMUN,, 1983 Table 1. Structural data for second- and third-stage graphite bifluorides, Non-intercalated C-atom sheet to C-atom sheet Hexagonal unit cell Compound parameters/nm separation/nm . ao = 0.2456(3) CnHF, {2 = PT seetbe 0.342(4) ao = 0,2456(3) C, HF, fa0 = 024 & 0.337(4) Intercalated C-atom shect to C-atom shect separation/nm Closest approach of F atoms to C-atom sheets/nm 0.602(4) 0.28(1) 0.607(4) 0.29(2) * For the 00/ data / = 2a, but observed Ak! reflexions require this doubling of ca. The stacking sequence of the carbon-atom sheets required to fit the observed line intensities is AB|BA|AB. third-stage salt has been preparcd by interaction of the second- stage salt with graphite in the presence of HF, Its composition lies within the range Cys-2,HF,. Both of these salts are dark grey solids (similar in appearance to graphite itself). They give sharp X-ray powder data to high Bragg angles and the analysis of that data, summarized in Table 1, indicates that the F atoms of the guest species (assumed to be HF,,) are approximately mid-way between the enclosing carbon-atom sheets of the host. The distance between adjacent carbon-atom sheets with- out intercalant is 3.4 A in each salt, a value not significantly different from that observed in graphite itself (3.348 A).3 The identity of the d spacings of the AKO diffraction data in these salts with the same data for graphite indicates that the average C-C nearest-neighbour distances within the carbon-atom sheets have not changed significantly with the intercalation, The increased separation of the sheets which accompanies intercalation (ca. 2.6A) isa ppropriate for the accommodation of ions with the size of bifluoride (ca. 2.6 A). In accord with the increase in the number of positive holes in the valence band which the C,*HF,- formulations suggest, each of these salts is an ab plane electrical conductor superior to graphite itself, The ratios at 20 °C of the specific conductivity} relative to that of the pristine graphite from which the salts were made (F:ograpnite) are for second Stage, 1.5:1, and third stage, All efforts to prepare a first-stage bifluoride have failed. Indeed, as intercalation of fluorine proceeds beyond the second Stage, CL HF,, a first stage material, C,F, is observed to be in admixture with it. In order to define the permissible values for x the interaction has been carried out using gaseous HF and fluorine, thus slowing the reaction. From such studies the maximum value for x in first-stage C,F has been found to be between 5 and 6. At this and subsequent points in the reaction sequence the recovery of HF employed in the preparations is almost complete. It appears that the co-operative local- isation of charge, by placement of HF,~ on each side of a carbon-atom sheet, renders the first-stage bifluoride unstable with respect to C,F and HF. However spccies such as HF,~ may continue to have a role in the intercalation of fluorine which continues, over a period of two days, to yicld a material of approximate composition C,F. (The hydrogen content of ‘C.F’ could be as high as that required for C,,HF,-4F but LIF recovery and analytical duta indicate that it is probably less.) The first-stage C,F (5.4 > x > 1.9) solids are black. Some of their physical Properties are set out in Table 2. Although the X-ray diffraction data for C,F are not of the high quality found for the bifluorides they are indicative of hexagonal symmetry and the 4k0 d spacings have nearly the same values as in graphite itself over the full range of x values. The d t Measurements made in collaboration with Dr. T. E, Thompson of SRI International, Mento Park, CA, ° Table 2. Some physical propertics of C,F. Parameters of hexagonal Ratio unit cell ¢ A — = Resistivity C/F Q/nm cofnm e/(Q2 cm) 5.28 0.2456(3) 0.522(2) >8 x 10-3 3.67 0.2459 0.536 8 x 1078 3.20 0.2456 0.545 3 x 10° 2.70 0.2460 0.571 2x 10% 2.49 0.2457 0.590 1x 10° 2.42 0.2459 0.602 1x 10° 2.15 0.2468. 0.622 1x 10’ 1.94 0.2466 0.645 2 x 10? * Contains some C,,HF, which is highly conductive, spacings for the 00/ data indicate a smooth increase in q, (which increases sharply for x < 3). The 00/ data are few how- ever, usually only 001 and 002 being observed. Morcover the powder lines are rather broad. This suggests that the inter- carbon-atom sheet spacing varies somewhat, either as a consequence of occasional guest-guest abutment or variation in the guest concentration within a gallery. The resistivity increases by ten orders of magnitude from graphite to CLF. The latter is an insulator. For values of x >2.4, however, the conductivity is sufficient to permit electrochemica} reduction (using HF containing NaF as elecirolyte and graphite as counter electrode) to higher-stage bifluorides.§ Moreover, it has been possible to prepare C,F with x as small as 2.6 by electrochemical fluorination of graphite using HF containing NaF as electrolyte, PbF, with lead as the other electrode, and a driving potential of 3.0 V. Open circuit voltages of C.F against lead have been determined as follows (x in paren- theses): (3.7), 2.03; (2.7), 2.2; (2.5), 2.38 V, Although the d spacings for the AKO diffraction data for C,F (Table 2) are nearly the same as in graphite, the relative intensities are not. It appears that the carbon-atom sheets must eclipse one another {in the [001] direction} and that the F atoms must also eclipse carbon atoms. However, the relative weakness of 002 relative to 001 shows that the F atonis cannot be near the mid-point between the enclosing carbon-atom shects. Evidently each F atom makes a close approach to one carbon atom (on the ¢ axis), and the attachment to carbon does not cause that atom to move out of the plane defined by its neighbours. § Some (510%) C,_,F, identified by its 001 diffraction line, usually remains in the electrochemically reduced C,F. / ‘i The available X-ray diffraction data‘are insufficient to prove the coplanarity of the carbon atoms. If the carbon atoms linked to F atomis are not coplanar with their neighbours the carbon to carbon bond distances projected into the ab plane are required to be coin- cidentally the same as the in-plane pristine graphite distances, This is not the case** for either (CF), or high-temperature C,E, J. CHEM. SOC., CHEM. COMMUN., 1983 - Table 3. XPS binding energies for C,.,;F and comparison materials. Compound Cls,eV F Is, eV Reference C3.3F 284.6 686.6 Present work? - 288.5 C.F 290.4 689.3 3 (CF:), 291.8 689.1 7 LiF 684.9 7 Grapnile 234.2 7 ® The recording of the XPS spectra and the reduction of data were-. carried out by Dr. K. D. Bomben and Prof. W. L. Jolly. From the X-ray photoelectron spectroscopy (XPS) binding energies given in Table 3 the F ligands of C,F are seen to be intermediate in character between the F in LiF and F in Teflon, whereas the F ligands in C.F prepared by Watanabe and his co-workers® appear to be more akin to F in Teflon. Evidently the F in C,.,F is semi-ionic in character, yet the bulk of the carbon closely resembles carbon in graphite. The ease with which the F ligand can be moved electrochemically is compatible with the semi-ionic nature of the C-F bond in C,F. Transmission i.r. spectra obtained from a chip of C,.,F {made thin by grinding perpendicular to the [001] axis} showed a strong absorption between 1050 and 1150 cm. This agrees with the observations® of Lagow er al. for C,F and is lower in frequency than the C-F stretch noted by Watanabe and his co-workers* for (CF),, and (C,F),, (1219 and 1221 cm7?). Semi-ionic C-F bonding in C,F is also supported by the Jong (2.1—2.2 A) C-F bond lengths suggested by structural models for C,F, C;F, and C.F. In these packing models the carbon- and fluorine-atom van der Waal’s radii are taken as 1.67 and 1.33 A, respectively. The fluorine atoms are distribu- ted evenly over the carbon atoms of the planar eclipsed sheets and the C-F bond length varies to produce the observed c- spacing. It might be that the fluorine is attached to the carbon in three-centre four-electron F-C-F bonds. The bonding mol- ecular orbital for such a three-centre sct has the symmetry appropriate for 7-bonding to its graphite environment. Also, the non-bonding orbital has no carbon orbital component. The long, highly ionic C-F bonds, the pseudo-graphite in- 105 sheet bonding, and the marked increase in resistivity,** as the composition C.F is approached, are thus accounted for. The authors gratefully acknowledge the support of the Director, Office of Energy Research, Office of Basic Energy Sciences, Chemical Sciences Division of the U.S. Department of Energy. The i.r. studies were made in collaboration with Mr. M. P. Conrad and Prof. H. L. Strauss. Received, 20th August 1982; Com, 10/3 References 1 T. E. Thompson, E. M. McCarron, and N. Bartlett, Syazth. Met., 1981, 3, 255. 2 W. Riidorff and G. Ridorff, Z. Anorg. Allg. Chem., 1947, 253. 281. 3 Y. Kita, N. Watanabe, and Y. Fujii, J. Am. Chez. Soc., 1979, 101, 3832, W. Ridorff and G. Riidorff, Chem. Ber., 1947, 80, 417. R. Wyckoff, ‘Crystal Structures,” Wiley, 2nd edn., Vol. J, 1963, p. 26. . W. Riidorff (Z. Anorg. Allg. Chem., 1947, 254, 319) reported a first-stage hydrofluoride of composition C,,HF,.4HF, with an Te of 8.06 A, preparéd by interaction of graphite with liquid HF and F, in quartz apparatus. Such matcrials have never been observed in the present studies. It may be that the quartz. is an essential ingredient for the first-stage material observed by Riidorff. 7 ‘Handbook of X-Ray Photoclectron Spectroscopy,” ed. G. E. Muilenberg, Perkin-Elmer Corporation, Eden Prairie, Minne- sota 55344, 1978. : 8 R. J. Lagow, R. B. Badachhape, J. L. Wood, and J. L. Margrave, J. Chenr. Soc., Dalton Trans., 1974, 1268, > nA Wh oy ** Although the bonding three-centre CF, orbital provides for =- bonding to the three C-atom neighbours, it is unlikely that such a w-system would provide for the passage of a current carrier (hole or electron). The pair of electrons in the orbital, because of its o- bonding rolc in binding the two F ligands, is more localized than the z-pair of elcctrons of the precursor’graphite system. The CF. and attendant three carbon atoms, C,CF,, would thercfore act as insulating islands. Composition C;CF, (i.e. C.F) would be an insulator, a F&O ae. : ore. c-e bonds). i $